Modélisation cinématique et dynamique avancée du membre supérieur pour l’analyse clinique
163
0
0
Texte intégral
(2) . . . . . . . . . . . . . . . .
(3)
(4) . !" # $ # #
(5) % & # '"() *+",)-. *)
(6) * !*/ 0* 1
(7) ) 2"
(8) -!".
(9) .
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(17) 3) * )456 "*
(18) 3 #7 #8 3) * )456 4 9 & "
(19) ) 2*931!6 &3:" $; >"
(20) 6#? & "*
(21) 3 #7 4 9 & >"/
(22) $ > >
(23) )& 9 "
(24) $ # . %< 66 *=&=&
(25) ) " '# . %< 66 *= <)*
(26) 6)9< 34
(27) @9= <)'. %< 66 * 6
(28) 6) 6=#$=9
(29) '> . %8%=# #"
(30) '> . #4 34 *'%4.=
(31) 6)9<" <=" <'B.
(32) @
(33) *'%4.=# > 8 4"C
(34) ) =*A 1/*@'. %=# $ 6)'> .. "!!) * "!!) * A"1
(35) ") * A"1
(36) ") * A"1
(37) ") * A"1
(38) ") * A"1
(39) ") *. .
(40) !"#$%&'!(##()'*+, . -
(41) ! . ./ 0 )1
(42) ''*. 3 8! 6
(43) )*#
(44) 6 1
(45) 6)")
(46) . &? %< 66 *"1"$ . 3 8! 6
(47) )*#
(48) 6 6)* 6 ) "
(49) 6)" . &? %< 66 *%4 !! . 3 8! 6
(50) )*#
(51) 6 3
(52) )<0* . &? %< 66 *; 1"
(53) ;. &? "
(54) . 3) *;
(55) " 6 3 6 . . . >"3* ) & 3
(56) 9
(57) 6)D# "* $
(58) ". 3) * &? %< 66 *F?. >"3* ) & 3
(59) ) &"E *)0* 9
(60) *D#4" 6 & *A. 3) * &1 "%< 66 * #?$ . >"3* )
(61) ) @ . 3) * &? %< 66 *?$ ; .
(62) 6))*) 63
(63) 3 6%4"1"3 *)0* 6 )$ @0* 6. 3) * &1 "%< 66 * #?#; .
(64) 6))*) 63
(65) 3 6 ) 34
(66) 0* 6 " ""!)")
(67) . 3) * &? %< 66 * ?& . !") 1
(68) ) <1")
(69) )#
(70) ) 34 34
(71) $ @ *1"
(72) . 3) * &1 ? "%< 66 * 8&?# . .
(73) . >"3* ) 63
(74) 3 6 ) 34
(75) @ 6. 3) * &?>?& #. !") 1
(76) )$ @ . 3) * &? %< 66 *>?> . !") 1
(77) )#41 $341 . 3) * &1 #"
(78) >H. !") 1
(79) );%. 3) * &?"66"
(80) && . !") 1
(81) )
(82) <1")0* !") 1
(83) )&")4 1")0* 6. 3) * &? %< 66 *? BB #. !") 1
(84) )& 3"
(85) 0* . 3) * &? %< 66 *?$ 3) * &?F "
(86) 8# "* %. !") 1
(87) )%4960* . 3) * &? %< 66 *; @ 6 & . > 3
(88) 3 6 ) 34
(89) 0* 6 6 3) ) 6%4960* 6 )!) 6 3) * &? ? % /6 ") 63
(90) 3 6
(91) 6 9
(92) . 3) * &?$?; . % 9) 349
(93) . 3) * &?%?> . 3 *! * #41 %4960* 3)
(94) 0* . 3) * &?;?%; .
(95) 6))*)
(96) 6)" 34
(97) @ 9
(98) . 3) * &? %< 66 *#? . 3 *! * *%< 66") ) *3")
(99) . 3) * &? %< 66 * ?& ; .
(100) 6))*) 3
(101) 3 >
(102) "
(103) 35 )G 66*"
(104) 3 6. 3) * &??$ . .
(105) Abstract Keywords: Soft tissue artefact, multibody optimisation, human movement analysis, stereophotogrammetry, modelling, upper limb, Soft Tissue Artefact (STA) is one of the most important limitations when measuring upper limb kinematics through marker-based motion capture techniques, especially for the scapula. Multi Body Optimisation (MBO) has already been proposed to correct STA when measuring lower limb kinematics and can be easily adapted for upper limb. For this purpose, the joint kinematic constraints should be as anatomical as possible. The aim of this thesis was thus to define and validate an anatomical upper limb kinematic model that could be used both to correct STA through the use of MBO and for future musculoskeletal models developments. For this purpose, a model integrating closed loop models of the forearm and of the scapula belt have been developed, including a new anatomical-based model of the scapulothoracic joint. This model constrained the scapula plane to be tangent to an ellipsoid modelling the thorax. All these models were confronted to typical models extracted from the literature through cadaveric and in vivo intracortical pins studies. All models generated similar error when evaluating their ability to mimic the bones kinematics and to correct STA. However, the new forearm and scapulothoracic models were more interesting when considering further musculoskeletal developments: The forearm model allows considering both the ulna and the radius and the scapulothoracic model better represents the constraint existing between the thorax and the scapula. This thesis allowed developing a complete anatomical upper limb kinematic chain. Although the STA correction obtained was not as good as expected, the use of this approach for a future musculoskeletal models has been validated.. .
(106) Résumé Mots-clés: Artefact des tissus mous, optimisation multi-segmentaire, analyse du mouvement, stéréophotogrammétrie, modélisation, membre supérieur Les Artefacts de Tissus Mous (ATM) sont actuellement une des limitations principales pour la mesure du mouvement du membre supérieur avec les techniques actuelles d'analyse du mouvement. L’optimisation multi-segmentaire (OMS) a déjà prouvé son efficacité pour la mesure du mouvement du membre inférieur. Afin d’avoir la meilleure correction possible, il est nécessaire d’utiliser des modèles d’articulation proches de l’anatomie. L’objectif de cette thèse a donc été de développer et de valider un modèle du membre supérieur qui pourrait être utilisé pour la correction des ATM par OMS. De nouveaux modèles en boucle fermée de l’avant-bras et de la ceinture scapulaire ont ainsi été développés accompagnés d’un nouveau modèle de l’articulation scapulo-thoracique imposant à la scapula d’être tangente à un ellipsoïde modélisant le thorax. Ces nouveaux modèles ont été confrontés aux modèles courants de la littérature à travers une étude avec vis intra-corticales sur cadavre et in vivo sur sujets asymptomatiques. Des niveaux d’erreur similaires ont été observés pour tous les modèles quant à leur capacité de corriger les ATM et d’imiter la cinématique osseuse. Les nouveaux modèles semblent cependant beaucoup plus intéressants dans une perspective de développement d’un modèle musculosquelettique. En effet, le modèle d’avant-bras autorise à la fois d’avoir le mouvement du radius et de l’ulna tandis que le modèle scapulo-thoracique représente mieux la contrainte existant entre le thorax et la scapula. En résumé, cette thèse a permis de développer un modèle complet proche de l’anatomie du membre supérieur permettant de corriger les ATM en utilisant une OMS. Bien que la correction des ATM obtenue n’est pas aussi satisfaisante qu’espérée, l’utilisation de cette approche pour le développement de futurs modèles musculo-squelettique a été validée.. .
(107) Remerciements Je souhaite tout d’abord exprimer ma gratitude au Fonds National de la Recherche Luxembourgeois pour avoir financé l’ensemble de cette thèse. Je voulais aussi remercier Frans C.T Van der Helm et Mickael Begon d’avoir accepté d’évaluer cette thèse, ainsi que Fabien Lebœuf et Sylvain Brochard de faire partie du jury. A l’issue de ce travail, je suis convaincu que même si la thèse est parfois un travail solitaire elle ne peut pas être réalisée sans le soutien d’un grand nombre de personnes. J’ai eu l’occasion de pouvoir réaliser ce doctorat entre différents laboratoires, de Luxembourg à Lyon, en passant par Caen ou Montréal. J’ai ainsi pu partager cette aventure avec de nombreuses personnes qui sont venues enrichir ce travail de leurs expériences et de leurs connaissances. Sans eux, cette thèse ne serait sûrement pas ce qu’elle est aujourd’hui et je tenais à les en remercier. Je voulais donc tout d’abord remercier le Centre National de Rééducation Fonctionnelle et de Réadaptation du Luxembourg (CNRFR–Rehazenter) pour m’avoir accueilli au sein de sa structure pour réaliser l’ensemble de ce travail. Je pense tout particulièrement à l’équipe du laboratoire d’analyse du mouvement et de la posture qui m’a accompagné au jour le jour (Dr.Paul Filipetti, Dr.Frederic Chantraine, Dr.Elisabeth Kolanowski, Angelique Remacle et Celine Schreiber). Un remerciement un peu plus spécifique à Florent Moissenet qui m’a encadré et a toujours réussi à canaliser (et supporter) mon trop plein d’énergie. Son implication dans le laboratoire et sa motivation constante ont été dans les moteurs principaux de l’ensemble de cette thèse. Merci de m’avoir toujours impliqué dans les discussions relatives à la recherche clinique ou aux développements du laboratoire. J’ai pu grâce à toi me construire une solide connaissance du fonctionnement d’un laboratoire d’analyse du mouvement en clinique. Si je ne devais avoir qu’un regret dans cette thèse c’est de ne pas avoir pu travailler de manière plus importante avec toi. Mes remerciements vont également à ma directrice de thèse Laurence Chèze ainsi qu’à Raphaël Dumas pour m’avoir accueilli et encadré au sein du Laboratoire de Biomécanique et Mécanique des Chocs et de m’avoir suivi tout au long de cette thèse. La semaine mensuelle que j’ai pu passer à Lyon a sensiblement participé au bon déroulement de ce travail. Les discussions que nous avons pu avoir m’ont permis d’avancer tant scientifiquement que personnellement. Un grand merci à vous pour vos conseils, votre dynamisme, et votre disponibilité. Les expérimentations réalisées au cours de cette thèse n’auraient pas été possibles sans la participation de mon co-directeur de thèse, le Dr. Thierry Haumont, que je tiens aussi à remercier. Merci de m’avoir ouvert les portes du laboratoire d’anatomie du CHU de Caen et de m’avoir fait confiance et aidé pour l’ensemble de ces travaux. Un remerciement aussi au Dr. Jean-Pierre Pelage et son équipe pour avoir exceptionnellement accepté d’ouvrir leur service d’imagerie en dehors des horaires d’ouverture pour que nous puissions réaliser nos différentes acquisitions. .
(108) Finalement, merci à Mickael Begon pour m’avoir accueilli pendant mon stage de trois mois dans son laboratoire (S2M) à Montréal. J’ai ainsi pu confronter mes différents modèles à la réalité expérimentale mais aussi à la rigueur du froid québécois. Merci à toute son équipe (Diane, Benjamin, Pat, Ariane, Colombes, Fabien et tous les autres) qui a su m’accueillir très chaleureusement contrairement à la météo. Je remercie aussi tous les autres doctorants du LBMC avec qui j’ai eu l’occasion de travailler au cours de cette thèse (Xavier, Angèle, Cindy, Julien, Stéphane, Brice, Pascal, Sylvie, Vincent, Romain). Merci pour toutes les pauses café (‘La machine à café est l’endroit où la recherche avance’), les conseils et la bonne ambiance qui pouvait régner dans la salle doctorant grâce à vous. Je tenais aussi à remercier mes colocs luxembourgeois (Marion, Ivan, Nicolas, Mélanie, les Rémi et les Clément) pour tous les bons moments de vie au RO85. La vie au Luxembourg aurait été beaucoup plus ennuyeuse sans vous. D’une manière générale, je voulais remercier chaleureusement tous mes amis (Nico, Jérémie, Thibaut, Elodie, Sandra, Hugo, Alain, André, Justine et tous les autres) pour avoir été là dans tous les moments depuis toutes ces années. Je voudrais remercier plus particulièrement ceux qui ont toujours eu la porte ouverte pour m’accueillir lors de mes semaines lyonnaises. Merci à Calwin, Daniel, Bastien, Blandine, Maxime, Baptiste et Magalie qui ont su m’offrir l’hospitalité en toute occasion et me rappeler qu’on peut toujours compter sur ses amis. Pour finir, je voudrais adresser toute mon affection à l’ensemble de ma famille. Merci pour tout l’amour et le soutien sans faille qu’ils m’ont donné tout au long de ma vie et de mes études, des plus petites classes de maternelles jusqu’à la fin, aujourd’hui, de mon doctorat. Merci pour toutes ces innombrables choses que vous m’avez apportées, petites et grandes qui font ce que je suis aujourd’hui.. Merci à tous pour ces trois années qui sans vous n’aurez pas été les mêmes !. .
(109) . Table of contents Abstract ............................................................... ii Résumé................................................................ iii Remerciements .....................................................iv General introduction ............................................. 1 1. Preamble .............................................................................. 1 1.1. Motion Capture.................................................................................................. 2 1.2. Associated limitations ........................................................................................ 3. 2. Aim of the thesis .................................................................. 4 3. Document organisation ......................................................... 5. Correction methods and models available ............. 7 1. Correction methods for STA-related errors .......................... 8 1.1. Regression formulae ........................................................................................... 8 1.2. Double calibration.............................................................................................. 9 1.3. Optimisation method ........................................................................................10. 2. Existing models................................................................... 12 2.1. Shoulder ............................................................................................................13 2.1.1. Glenohumeral ........................................................................................................13 2.1.2. Shoulder girdle.......................................................................................................14 2.1.2.1. Equivalent mechanism: first approximation for modelling shoulder girdle......15 2.1.2.2. Anatomical representation: Open loop vs Closed loop ....................................16 Open loop model: Considering the different bony structures of the shoulder girdle 16 Closed loop model: Integration of the scapulothoracic joint ....................................17. 2.2. Forearm model: integration of a real pronosupination .....................................18 2.2.1. Simple model: Integration of the ulna and radius .................................................20 2.2.2. Integration of a more realistic joint kinematic model: Lemay, Pennestri and Weinberg models .............................................................................................................20 .
(110) Table of contents 2.2.3. Integration of supplementary degrees of freedom. .................................................23. 3. Geometric parameters ........................................................ 24 3.1. Prediction methods ...........................................................................................24 3.2. Regression and scaling methods ........................................................................25 3.3. Functional methods ..........................................................................................26 3.4. Imaging techniques ...........................................................................................26. 4. Validation procedures ......................................................... 28 4.1. Imaging techniques ...........................................................................................28 4.2. Palpation methods ............................................................................................29 4.3. Intracortical pins ...............................................................................................30. 5. Thesis choices: .................................................................... 31. Theoretical background ....................................... 32 1. Natural coordinates and non-orthogonal coordinate system 32 2. Multibody system with natural coordinates ........................ 34 3. Upper limb model ............................................................... 35 3.1. Kinematics calculation ......................................................................................38 3.2. Virtual marker and vector definition ................................................................39. 4. Multibody optimisation ...................................................... 40 4.1. Motor constraints..............................................................................................40 4.2. Rigid body constraints ......................................................................................41 4.3. Kinematic constraints .......................................................................................42 4.3.1. Scapulothoracic joint .............................................................................................42 4.3.1.1. Model tangent to an ellipsoid..........................................................................42 4.3.1.2. Contact point model .......................................................................................45 4.3.2. Clavicle ..................................................................................................................47 4.3.3. Glenohumeral joint ................................................................................................48 4.3.3.1. Case 1: Spherical joint ....................................................................................48 4.3.3.2. Case 2: Constant length ..................................................................................49 4.3.4. Forearm .................................................................................................................49 vii .
(111) Table of contents 4.3.4.1. Humeroradial joint ..........................................................................................49 4.3.4.2. Humeroulnar joint ...........................................................................................50 4.3.4.3. Distal radioulnar joint.....................................................................................51 4.3.5. Wrist .....................................................................................................................51. 4.4. Final formulation ..............................................................................................52. Cadaveric study .................................................. 53 1. Experimental protocol ........................................................ 54 1.1. Cadaver preparation .........................................................................................54 1.2. Motion capture protocol ...................................................................................57 1.3. CT-scan .............................................................................................................58 1.4. Model generation ..............................................................................................59 1.5. Models comparison ............................................................................................61 1.6. Contact point ....................................................................................................62 1.6.1. Definition ...............................................................................................................62 1.6.2. Equation ................................................................................................................64. 2. Results ................................................................................ 65 2.1. Scapulothoracic joint ........................................................................................65 2.1.1. Kinematics .............................................................................................................65 2.1.1.1. Upward-downward rotation ............................................................................69 2.1.1.2. Protraction-retraction .....................................................................................69 2.1.1.3. Anterior-posterior tilt......................................................................................69 2.1.2. Contact point ........................................................................................................70. 2.2. Forearm.............................................................................................................72. 3. Discussion ........................................................................... 73 3.1. Scapulothoracic model ......................................................................................73 3.1.1. Kinematics .............................................................................................................73 3.1.2. Contact point ........................................................................................................74 3.1.3. Ellipsoid definition.................................................................................................75. 3.2. Forearm model ..................................................................................................77. 4. Limits ................................................................................. 77 viii .
(112) Table of contents 5. Conclusion .......................................................................... 78. In vivo intracortical experimentation .................. 79 1. Validation data ................................................................... 81 1.1. Intracortical pins ...............................................................................................81 1.2. Comparison method ..........................................................................................83. 2. Study 1: Ability of the different MBO models to mimic the bones kinematics ..................................................................... 84 2.1. Methods ............................................................................................................84 2.2. Results ..............................................................................................................85 2.2.1. Anterior-posterior tilt ............................................................................................87 2.2.2. Protraction-retraction ............................................................................................87 2.2.3. Upward-downward rotation ...................................................................................88. 2.3. Discussion .........................................................................................................88. 3. Study 2: Ability of the different MBO models to correct Soft Tissue Artefact ....................................................................... 88 3.1. Method ..............................................................................................................88 3.2. Results: using ISB motor constraints................................................................89 3.2.1. Upward-downward rotation ...................................................................................91 3.2.2. Protraction-retraction ............................................................................................91 3.2.3. Anterior-posterior tilt ............................................................................................92. 3.3. Results: using Jackson et al. motor constraints ................................................92 3.3.1. Upward-downward rotation ...................................................................................92 3.3.2. Protraction-retraction ............................................................................................92 3.3.3. Anterior-posterior tilt ............................................................................................93. 3.4. Discussion .........................................................................................................95. 4. Study 3: Simpler STA correction method ........................... 96 4.1. The neglected-Degree of Freedom method........................................................97 4.1.1. Equation ................................................................................................................97 4.1.2. Experimentation: definition of different marker sets .............................................98. 4.2. Results ..............................................................................................................99 ix .
(113) Table of contents 4.2.1. Anterior-posterior tilt ............................................................................................99 4.2.2. Upward-downward rotation ...................................................................................99 4.2.3. Protraction-retraction .......................................................................................... 100. 4.3. Discussion ....................................................................................................... 102. 5. Limits ............................................................................... 103 6. Conclusion ........................................................................ 104. General conclusion ............................................ 106 1. Thesis outcomes ................................................................ 106 1.1. MultiBody Optimisation as a tool for correcting Soft Tissue Artefact........... 106 1.2. Kinematic chain model development .............................................................. 106 1.3. Models validation: Intracortical pins .............................................................. 107 1.4. Conclusion ....................................................................................................... 107. 2. Perspectives ...................................................................... 108. Bibliography ...................................................... 113 List of figures .................................................... 122 List of tables ..................................................... 127 Appendix ........................................................... 128. x .
(114) . General introduction 1.Preamble The main upper limb functionality is to set the end-effector (i.e., the hand) position and orientation in space, with or without load. In order to achieve this task, a compromise between mobility and stability, strength and loading has to constantly be found. Since the upper limb is often required for repetitive tasks, these constraints can become traumatic and induce osteoarticular or musculo-tendinous disorders (Anglin and Wyss, 2000). The upper limb is also highly impacted by neurological pathologies such as cerebral palsy, hemiplegia, paraplegia or tetraplegia. In that case, the movements of the patient is faced with different motor control impairments (e.g., muscle lack of selectivity, weakness, spasticity). It results in important deficiencies in performance of daily activities and impacts thus directly the quality of life and the independency of the patient. In order to choose an adapted treatment for each patient, an evaluation of the different impairments of their upper limbs is essential. The assessment of these patients is mainly based on a clinical evaluation (e.g., muscle strength, spasticity, range of motion). Sometimes, this evaluation is completed by clinical scales based on functional movements (e.g., Melbourne Assessment of Unilateral Upper Limb Function. (Bourke-Taylor, 2003), Quality of Upper. Extremity Skill Test (DeMatteo et al., 1992)). However, all these evaluations tend to stay more qualitative than quantitative and can be subjective as they are based on the observation of an assessor. In addition, some tasks might not be sensitive enough to detect changes that would be relevant clinically (Sätilä et al., 2006). Consequently, there is a lack of quantitative analysis in the evaluation of the upper limb in a clinical context. Some quantitative analysis can be found through the literature and focus commonly on two elements: spatiotemporal characteristics and joint kinematics (Jaspers et al., 2009). The spatiotemporal characteristic are generally associated to the time taken by the subject to perform a specific functional task corresponding to daily activities (e.g., reach and grasp, hand to mouth, hand to back…). This allows having a global quantitative evaluation of the upper limb. However, in order to be able to identify the different compensatory movements or patterns that can exist, and then allow for more precise diagnosis, there is a need to focus on each joint separately. For this purpose, analysing the different joint kinematics and dynamics is useful. It can allow relating the different movement impairment characteristics to the underlying pathology and define different classifications based on these findings. However, the kinematic and dynamic explorations of the whole upper limb during a movement remains challenging. The diagnosis and treatment prescription are thus limited. The main difficulty is located at the shoulder, composed of five high mobility joints and a complex polyarticular muscles layout. A better understanding of the shoulder complex functioning during a movement could allow significant improvements in terms of diagnosis and treatment. .
(115) General introduction Motion analysis is a solution that can be used to reach this goal. Its principle is to determine the position and orientation of different body segments in space. This allows obtaining derived information such as the kinematics of each joint by the knowledge of the position and orientation of the anatomical coordinates systems embedded in the body segments. This methodology is already widely used to study the lower limb kinematics when exploring gait (Cappozzo et al., 2005; McGinley et al., 2009; Woltring, 1994). Gait is an efficient movement allowing the energy loss minimisation during the transport of the human body. It is an automatic and repeatable movement. Consequently, the definition and the comparison with normative data is relatively easy and adaptable in a clinical context. Unlikely, due to the high range of motion of the different joints constituting the upper limb, several strategies can arise for a similar task between subjects (Rau et al., 2000). It results that the definition of normative data, classification or the comparison between subjects is much more challenging than for lower limb in a clinical context. Beyond these problems linked to the complexity of the upper limb, the evaluation of the upper limb remains difficult due to technical limitations described below.. 1.1.Motion Capture One of the most common techniques in motion analysis is stereophotogrammetry. Stereophotogrammetry allows determining the spatial position of an object based on multiple cameras recordings. The main technology used is based on optoelectronic cameras which allow reconstructing the 3D position of passive cutaneous reflective markers bouncing back an infrared light emitted by the cameras. Using this technology, the precision for a marker position can be below 1 mm (Chiari et al., 2005). In order to use stereophotogrammetry, the position and the orientation of each camera must be known. A calibration procedure is thus needed. This procedure must be done carefully as if it is done poorly, errors on the markers position can arise. Basically, the calibration results in the definition of the laboratory coordinate system (i.e., the inertial coordinate system or ICS) and the position and orientation of each camera in this coordinate system. Once this calibration has been performed, markers glued on the skin can be used to define the Segment Coordinate System (SCS) of the associated body segments. Indeed, these body segments are generally considered as non-deformable as the global interest is their position in space. Different approaches have been proposed to determine SCSs. On one hand, SCSs can be geometrically defined directly through a set of cutaneous markers positioned on palpated anatomical landmarks. These landmarks are generally associated to bony prominences that can be palpated in a repeatable manner (Van Sint Jan, 2007). The axes, defined using these markers, must then approximate as precisely as possible the different anatomical planes (i.e., frontal, transverse and sagittal planes) of the segments. On the other hand, SCSs can be obtained through a Calibrated Anatomical System Technique (CAST) (Cappozzo et al., 2005). Using CAST, a cluster of markers is positioned on each body segment to define a local technical coordinate. 2 .
(116) General introduction system. These clusters can be placed directly on the skin, on an elastic band or on a rigid plate attached to the body segment. In order to determine the SCSs, each anatomical landmark position is then defined during a static posture data collect and assumed fixed in this technical coordinate system. Thus, it is possible to compute the position of the SCS during the complete motion. The main interest of the CAST method is the possibility to choose the position of the cluster considering other constraints (e.g., the markers forming the cluster can be for example positioned in order to avoid any marker visual loss during acquisition or to minimise their movement with respect to the underlying bones). Finally, once each SCS has been defined, joint kinematics can be computed through Euler angles definition (Wu et al., 2002).. 1.2.Associated limitations Even with a technology able to determine the position of reflective markers with a precision under a millimeter, some limits exist when using stereophotogrammetry. A first limit is the palpation technique used for determining the different anatomical landmarks. Indeed, their positioning impacts directly on the SCS and thus kinematic calculations. Consequently, if the intra-/inter-session and/or operator repeatability is low due to a poor palpation technique, differences can arise in angles patterns without any change of the subject kinematics. This can be illustrated by a phenomenon called cross talk phenomenon. If the main rotation axis (e.g., knee flexion-extension axis) is poorly defined, the associated rotation will be reverberated in another rotation axis (e.g., knee abduction-adduction) (Della Croce et al., 2005; Piazza and Cavanagh, 2000). Another limit is called Soft Tissue Artefact (STA). STA arises from the relative movements between the cutaneous markers and the underlying bone (Leardini et al., 2005). Indeed, even if theoretically each body segment is considered as a rigid segment, this one is composed of the bony rigid part (i.e., our centre of interest) covered by deformable tissues (i.e., muscles, skin, fat tissues) that move during a movement. As a result, a difference exists between the position of the cutaneous markers and that of the associated anatomical landmarks on the bone. Consequently, as the position of the different body segments is determined directly by external landmarks, a difference can appear between the constructed SCS and the true bone poses resulting in kinematic errors. In the context of the upper limb, and particularly for the scapula, STA can be important and limit the use of motion capture. Indeed, in this case, the position error between the cutaneous markers and the anatomical landmarks can be up to 8 cm (Matsui et al., 2006)(Figure 1). . 3 .
(117) General introduction. Figure 1 : Difference between the position of some anatomical landmarks (i.e., TSp, AIp palpated in the current position) and associated cutaneous markers (i.e., TS, AI palpated in the resting position) (Senk and Chèze, 2010). As a result, STA can impact significantly the kinematics with errors up to 35° in internal-external rotation of the shoulder (Cutti et al., 2008). Correcting STA-related errors is then crucial to allow the use of motion analysis in a clinical environment and to obtain reliable data for the upper limb kinematics. As a result, the scapula kinematics is currently often neglected in clinical study. In other word, the kinematics is alternaltively computed for a virtual humero-thoracic joint.. 2.Aim of the thesis As suggested, the use of stererophotogrammetry as a tool for measuring the movement in a clinical context seems relevant and could allow obtaining supplementary information for patient diagnosis and follow-up. However, this technic is strongly limited by STA. Thus, to propose and develop methods allowing to correct STA-related errors in a clinical context seems crucial. As it will be presented in Chapter 1, the correction method chosen in this thesis was based on the development of a MultiBody Optimisation (MBO) model of the upper limb. This kind of model requires the development of precise and physiological joint kinematic models. As a result, this work was also integrated in a larger project of the Laboratoire de Biomécanique et Mécanique des Chocs (LBMC-IFSTTAR, Université Claude Bernard Lyon 1) that aims developing a full musculoskeletal model of the upper limb (i.e., integrating the dynamics and muscle behavior). Such a model would allow obtaining musculo-tendon and joint reaction forces from a simple motion analysis in addition to the classical kinematics (Garner and Pandy, 2001). However, before obtaining correct dynamics and muscle behaviour, there is a need to assess the correct bones kinematics. Without precise and physiological kinematic models, it is difficult to obtain a 4 .
(118) General introduction reliable bones kinematics. In order to be able to use the different kinematic models developed as a base for further developments or in a clinical context, validation is needed. This project aims thus not only to develop kinematic models but also to confront them to required validation process. In short, this thesis aims at the definition and the validation of a correct upper limb kinematic model that could be used both to correct STA-related errors through the use of MBO and for future musculoskeletal model developments.. 3.Document organisation Beside this introduction, this document is composed of four different chapters and a general conclusion. Chapter 1 - Correction methods and models available – In order to be able to use MultiBody Optimisation (MBO) for correcting Soft Tissue Artefact (STA), it is necessary to define a proper upper limb kinematic model. However, before being interested in the development of such a model, it must be understood why this approach has been chosen. Correcting STA can be done through different methods and, in order to appreciate the different choices made in this thesis, it is necessary to determine their advantages and drawbacks. This chapter aims thus answer three main questions: 1) Which correction method seems the most adapted for correcting STA in upper limb? 2) Considering the chosen method (i.e., MBO), which associated kinematic models, in terms of modelling and geometry, should be used in order to obtain the best results? 3) And finally, how these methods and models can be validated? This chapter is organised as a review of literature of the methods and models available. Each subchapter justifies the different technical choices that have been made further in the project for each of the following elements: STA correction method, kinematic models and validation tools. Chapter 2 - Theoretical background – The previous chapter introduced the MBO approach which will be used to correct STA. Such an approach requires: 1) Defining the parameters used to position the body segments of the upper limb kinematic chain, 2) Establishing the optimisation procedure, 3) Defining the kinematic constraints of each joint.. 5 .
(119) General introduction The parameters used to position the body segments will yield the design variables of the optimisation. They also correspond to the different SCSs required to perform the kinematic computation. The aim of this chapter is thus to develop and to present all the different equations used in our upper limb model integrating the complete shoulder complex and the movements existing between the ulna and the radius. The last two chapters confront our MBO model to validation data. The abilities of the model to mimic the bone kinematics and to correct STA-related errors were then evaluated. Chapter 3 - Cadaveric intracortical experiment – The aim of this study was to evaluate the ability of our model to mimic the bone kinematics. The first experiment was conducted at the Anatomical laboratory of Caen and consisted in a cadaveric study. This experimentation aimed at: 1) Measuring precisely the bone kinematics of the different body segments in order to validate the kinematic models and the ability of MBO to mimic human motion. 2) Obtaining geometric parameters of the complete upper limb kinematic chain and comparing the results obtained when a method with parameters scaled from a literature model is used. This experimentation was then constituted of two parts. The first part was a motion analysis with intracortical pins inserted the cadaver upper limb and thorax in order to obtain validation data for both the shoulder complex and the forearm kinematic models. In the second part, a CT-scan was performed after having frozen the cadaver in anatomical position in order to extract geometric parameters. It was then possible to obtain the real geometry of the model and to use it to construct the MBO model. Chapter 4 - In-vivo intracortical experiment – Thanks to the collaboration established with the S2M laboratory (Montreal, Canada), it was possible to access intracortical pins in vivo data. Due to the invasiveness of the experiment, only the shoulder complex was studied. The main interest of these data, compared to the ones obtained with the cadaveric study, is that it was both possible to access the bones kinematics in active movements and to evaluate the effect of STA. These data allowed to: 1) Evaluate more precisely the ability of the different scapulothoracic kinematic models to mimic the bone kinematics on two living subjects. 2) Evaluate the ability of different MBO models to correct STA-related errors. 3) Develop and evaluate a new simpler STA correction method based on the elimination of the influence of a marker on one specific degree of freedom.. 6 .
(120) . Chapter 1 Correction methods and models available In order to be able to use MultiBody Optimisation (MBO) for correcting the Soft Tissue Artefact (STA), it is necessary to define a proper upper limb kinematic model. However, before being interested in the development of such a model, it must be understood why this approach has been chosen. Correcting STA can be done through different methods and, in order to appreciate the different choices made during this project, it is necessary to determine their advantages and drawbacks. This chapter aims thus answer three main questions: - Which correction method seems the most adapted for correcting STA in upper limb? - Considering the chosen method (i.e., MBO), which associated joint models, in terms of modelling and geometry, should be used in order to obtain the best results? - And finally, how these methods and models can be validated? This chapter is organised as a review of literature of the methods and models available. Each subchapter justifies the different technical choices that have been made further in the project for each of the following elements: STA correction method, kinematic models and validation tools.. Contents 1. Correction methods for STA-related errors .......................... 8 1.1. Regression formulae ........................................................................................... 8 1.2. Double calibration .............................................................................................. 9 1.3. Optimisation method ........................................................................................10. 2. Existing models................................................................... 12 2.1. Shoulder ............................................................................................................13 2.2. Forearm model: integration of a real pronosupination .....................................18. 3. Geometric parameters ........................................................ 24 3.1. Prediction methods ...........................................................................................24 3.2. Regression and scaling methods ........................................................................25 + .
(121) Correction methods and models available 3.3. Functional methods ..........................................................................................26 3.4. Imaging techniques ...........................................................................................26. 4. Validation procedures ......................................................... 28 4.1. Imaging techniques ...........................................................................................28 4.2. Palpation methods ............................................................................................29 4.3. Intracortical pins ...............................................................................................30. 5. Thesis choices: .................................................................... 31 . 1.Correction methods for STA-related errors In order to be able to correct STA-related errors, different approaches and methods have been proposed in the literature for the upper limb.. 1.1.Regression formulae One of the most common correction methods consists in using regression formulae. A regression formula determines the supposed true position of the anatomical landmarks based on the position of the cutaneous markers as a function of specific variables (e.g., joint angles). It can also allow defining the position and orientation of a body segment based on the position and orientation of another body segment. These formulae are commonly based on multiple static acquisitions where the anatomical landmarks are palpated and their positions expressed relative to the cutaneous markers. The postures used for these multiple static acquisitions are selected to allow defining the precise position and orientation of the different body segments. Linear regression parameters can then be extracted and the regression formulae defined from this data base. In this sense, DeGroot et al. (2001) defined a regression formula for the scapula rhythm and the clavicle positioning as a function of the humerus elevation and the plane of elevation. This formula was based on 23 positions spread on six elevations (i.e., 0, 30°, 60°, 90°, 120°, 150°) and 4 planes of elevation (i.e., 30°, 60°, 90°, 120°). More recently, Lempereur et al. (2010a) quantified the STArelated errors during elevation to define a correction factor for the three rotation degrees of freedom (DoF) of the scapula (i.e, protraction-retraction, anterior-posterior tilt and upwarddownward rotation) as a function of the humeral elevation. These new formulae allowed defining the real position of the scapula based on the collected position with STA-related errors. However, as it has been shown by MCClure et al. (2011), the position in axial rotation of the arm impacts the kinematics of the scapula. Therefore, it can be supposed that even if these regression formulae seem working on simple movements, their validity for more complex ones can be questioned as axial rotation of the humerus was not considered in their establishment. Thus, Grewal & 8 .
(122) Correction methods and models available Dickerson (2013) and Xu et al. (2014) proposed a regression formula integrating the internal and external rotation position of the humerus allowing its use on a wider variety of movements. Anyway, a daily clinical use of these methods remains difficult due to the inter-subject variability (McClure et al., 2011). This seems even more challenging when considering pathological populations (Anglin and Wyss, 2000). As a result, regression formulae do not seem adapted to our requirements due to the important variability of the subjects (e.g., healthy vs. pathological) and because there are limited to simple movements. However, they could be used as a first approximation for clinical studies. Integrating subject-specific measurements in order to personalise the methods could also be interesting and may allow a better adaptability of these formulae. Such an approach is detailed is the next paragraph.. 1.2.Double calibration Double calibration has been initially developed for the lower limb and more precisely when exploring the knee kinematics (Cappello et al., 1997). The basic principle of this method is to measure the position of both different cutaneous markers and anatomical landmarks or bone specific directions relative to a technical coordinate system (like in CAST method) in two extreme positions of the studied joint (e.g., full extension and full flexion for the knee) on the analysed subject. The joint positions used can also be limited to the range of motion observed during the investigated movement. Assuming that the cluster deformation is coherent during the full movement, this method is used to interpolate the bone kinematics based on the corrected position of the clusters all along the movement. It thus aims obtaining a better kinematics of the studied joint by considering the possibility of both a deformation and rigid displacement of the clusters during the movement. This method has been validated through a fluoroscopy study for lower limb (Cappello et al., 2005) and adapted and validated for the scapula during an elevation movement through palpations during static postures using a scapula locator (Brochard et al., 2011) (Figure 2).. Figure 2 : Double calibration for the scapula during elevation. 9 .
(123) Correction methods and models available However, double calibration has been mainly validated on movements with one main degree of freedom such as knee flexion (e.g., during walking) or arm elevation. For the knee, this method seems perfectly adapted as the knee is a joint with one main independent degree of freedom (i.e., around the flexion-extension axis). However, the upper limb is composed of joints with motion occurring in all different anatomical planes. Consequently, though this method seems decreasing STA-related errors, its use for the upper limb seems compromised. Moreover, the definition of the extreme positions for a complex movement which can be found in upper limb appears more difficult.. 1.3.Optimisation method Methods based only on markers position have also been proposed for correcting STA-related errors. They can be classified as optimisation methods. Two main methods can be considered and opposed to the direct method (i.e., method that refers to the classical method which determines the position of the body segments directly from the position of the different cutaneous markers without any correction): the single-body (segmental) optimisation method and the MBO method. In both of these methods, the deformation of the clusters of markers is minimised. As it has been shown that the direct method can be impacted by STA (Cappozzo et al., 2005), the segmental optimisation considers each body segment separately. In that condition, the deformation of the clusters, composed of a minimum of three non collinear markers, is minimised for each frame using a least squared method (Cheze et al., 1995). As a result, it is possible to minimise STA-related errors and to obtain a better bone pose. However, by considering each segment separately, non-physiological dislocations may appear at the joints. Indeed, the movement of the clusters of markers, due to STA, can be composed of different deformations (e.g., stretch, homothety) and rigid body motions (i.e., translation, rotation) (Dumas et al., 2014). Unfortunately, while segmental optimisation can correct the deformations, it cannot correct the rigid body motions. As a result, a dislocation appears when the rigid motion component of the STA is not compensated. MBO can be interesting to avoid this dislocation problem. Instead of considering each body segment separately, this method considers the limb as a kinematic chain. This kinematic chain is composed of the different segments supposed rigid and connected with joints (Dempster, 1965). These joints can be simplified as basic mechanical links such as spherical, hinge or universal joints (Andersen et al., 2009; Duprey et al., 2010). The position and orientation of all segments are then optimised together under kinematic and rigid body constraints in order to minimise the sum of the squared distances between measured and model-determined markers positions (Figure 3). By taking into account all the different segments and their associated markers, this method aims thus compensating for both STA components: rigid body motions and deformations.. 10 .
(124) Correction methods and models available. Figure 3: Multibody optimisation can be used to minimise the squared distance between measured and model-determined markers positions (Andersen et al., 2009). However, the results obtained with MBO strongly depend on the mechanical links chosen for each joint (Duprey et al., 2010). As an example, it would be impossible to obtain any rotation at the knee, which is a part of its stabilisation mechanism, if the joint is modelled by a hinge. As a result, the model for each joint should be carefully chosen. While simple mechanical mechanisms are usually employed for joint modelling, one of the main interests of MBO is the possibility to use more complex ones such as parallel mechanisms. Parallel mechanisms consist of different mechanical constraints used for the same joint (in parallel) in the purpose to obtain a model closer to the anatomy. For example, the knee can be considered as two spheres (i.e., the two femur condyles) in contact with a plane (i.e., the tibial plateau) and guided by different connecting rods corresponding to the knee ligaments (i.e., lateral and cruciate ligaments) (Figure 4). This kind of mechanism has also been adapted for the ankle where the contacts between the tibia and the talus can be considered as 3 sphere-on-plane contacts or more simply a spherical joint guided by two connecting rods corresponding to the medial and lateral ligaments (Figure 4). It has been shown (Duprey et al., 2010) that the use of such physiological models allows obtaining a better correction of the kinematics. As a result, the coupled use of MBO and physiological kinematic models of the different joints may improve the quality of the results obtained with motion capture analysis and has been chosen in this thesis as a STA-related errors correction method. One of the main works will thus remain on the adaptation of this method for the upper limb. This can be done through the choice of a set of considered body segments and the selection of different joint kinematic models defining the complete upper limb kinematic chain that should be used in the MBO process. A non-exhaustive list of the models proposed in the literature is given in the following paragraphs.. 11 .
(125) Correction methods and models available. Figure 4: Different ferent parallel mechanisms for multibody optimisation forr the knee and the ankle (Moissenet, 2011) 201. 2.Existing models odels The most common upper limb b model integrates the thorax, the arm, ar the forearm and the hand cted by the shoulder, the elbow and the t wrist modelled as hinge or as rigid body segments connected. spherical joints (Ambrósio et al., 2011) (Figure 5).. Figure 5 : Common upper limb model (adapted from Maurel & Thalmann 1999). While such a model can be used as a first and simplified model, two important issues must be reported. Firstly, the shoulder girdle, composed of the clavicule, scapula and thorax with their three associated joints, is not modelled. Indeed, in such a model, the movement of the shoulder girdle is only considered through the glenohumeral (GH) joint which is assumed fixed in the thorax SCS, while the shoulder girdle movement can modify completely the position of this joint. Secondly, the forearm model does not consider the two forearm bones (i.e., the ulna and the radius), simplifying the associated pronosupination movement between them as a supplementary DoF at the elbow or the wrist joint (Schmidt et al., 1999). However, these two mechanisms have a key role to ensure the important range of motion of the upper limb and the precise positioning of the hand. In this sense, the proper integration in an upper limb model of the shoulder girdle 12 .
(126) Correction methods and models available (Gamage and Lasenby, 2002) and of the pronosupination movement should be considered as essential for clinical analysis. Advanced scapular and forearm models should allow to better replicate human anatomy and thus improve the quality of the correction of STA-related errors, as it has been shown previously for the lower limb (Duprey et al., 2010).. 2.1.Shoulder 2.1.1.Glenohumeral The GH joint is composed of the glenoid fossa (or glenoid cavity) and the humeral head connecting the scapula to the humerus (Figure 6). The two contact surfaces are spherical resulting in a joint allowing three rotations as DoF. This joint is highly instable due to its bony structures allowing a high range of motion (i.e., abduction 150°-180°, flexion 180°, extension 45-60° and internal external rotation 90°). Indeed, contrary to the hip where the concave surfaces (i.e., acetabulum) surround well the convex surfaces (i.e., femoral head), the glenoid represents a small surface. Consequently, the stability of the joint is mainly assumed by the ligaments surrounding it and by the coaptation effect of the muscles of the rotator cuff which ensure that the humeral head remains in contact with the glenoid fossa.. Figure 6 : Glenohumeral joint illustration (adapted from Netter (2006)). GH joint is commonly modelled as a spherical joint in musculoskeletal models (Engin and Tümer, 1989; Garner and Pandy, 1999; Högfors et al., 1991; Maurel and Thalmann, 1999; van der Helm, 1994). Indeed, in first approximation, this mechanical joint represents perfectly the movement existing between the scapula and the humerus (i.e., the gliding of the spherical humeral head in the concave glenoid fossa). However, as these two elements can have different radii, their movement has also been described as ‘the movement of a ball on a seal nose’ (Hill et al., 2008) suggesting both rotation and translation movements. In practice, while it has been shown through cadaveric studies (Harryman et al., 1990; Kelkar et al., 2001) and a fluoroscopy study (Dal Maso et al., 2014; Hill et al., 2008) that the translation during the 60° first degrees of flexion can reach 13 .
(127) Correction methods and models available 3mm, and up to 5mm for a complete flexion, the translation movement is commonly neglected. Consequently, it could be interesting to introduce translation in the modelled GH joint. Nevertheless, the use of a six DoF joint could almost be considered as an error or at least a misuse of the MBO method. Indeed, the main interest of this method is to use the kinematic constraints of the different joints to minimise the STA-related errors. Not using any constraint at the GH joint joint (i.e., by the definition of six DoF) would correspond to perform a MBO on two kinematic chains separately: one on the shoulder girdle complex (i.e., thorax, clavicle, scapula), and one on the remaining structures of the upper limb. A solution to avoid such a situation could be to allow translation but only on a specific range. The kinematics of the GH joint can thus be considered as two spheres of different radii rolling with one another which can be modelled as a rigid link between the two rotation centres (i.e., glenoid fossa and humeral head centres) such as proposed by El Habachi et. al (2015a). Such a model allows taking into account a translation between the scapula and the humerus in a physiological range, while remaining suitable for its use in a MBO procedure. Another approach is to consider ‘soft’ constraints or, in other words, to manage the GH spherical constraint with a penalty-based method (Charbonnier et al., 2014). 2.1.2.Shoulder girdle While several GH models have been discussed in the previous paragraph, the main differences when modelling the shoulder are due to the representation of the shoulder girdle composed by the scapula, the clavicle and the thorax. This closed chain allows positioning the GH joint in a correct manner and forming with this joint the shoulder complex. Shoulder girdle is composed of 2 true synovial joints: the sternoclavicular (SC) and acromioclavicular (AC) joints, completed by the scapulohoracic (ST) joint which is not considered as a true joint (i.e., no articular structure involved). The SC joint (Figure 7) is the joint connecting the clavicle to the thorax through the manubrium and the cartilage of the first rib. This joint is commonly assimilated to a saddle joint as the two articular surfaces have a concave and convex torus shape. The torus shape could be assimilated as an inner tube. Consequently, mainly two DoF are allowed through this joint in the vertical and horizontal planes of the clavicle. The AC joint (Figure 7) is composed of the scapula acromion and the lateral part of the clavicle. This joint allows rotation movements in the three planes but does not allow translation. This joint, due to small articular surfaces, is highly instable and prone to luxation. Its stability is mainly assumed by two external ligaments, the trapezoid and the conoid ligaments, positioned between the clavicle and the coracoid process.. 14 .
(128) Correction methods and models available. Figure 7 : Sternoclavicular and acromioclavicular joints (adapted from Netter (2006)). No articular structure is involved in the ST joint (Figure 8), this is why it is commonly not considered as a true joint. However, the movement of the scapula is constrained by the surrounding muscles to glide over the thorax above the serratus anterior and the subscapularis muscles.. . . . Figure 8 : ScapuloThoracic joint in a coronal view (T3-T4) (adapted from Netter (2006)). 2.1.2.1.Equivalent mechanism: first approximation for modelling shoulder girdle The shoulder girdle can be firstly modelled as an equivalent mechanism representing the movements of the scapula and positioning correctly the GH joint (which, as it has been shown, is commonly modelled as a spherical joint). Different models of this kind can be found in the literature. Lenarcic and Umek (1994) and Malek et al. (Malek et al., 2006) used a simple universal joint while Yang (2005) used 2 prismatic joints. A more complex model was also proposed by Klopcar and Lenarcic (2006; 1999) where a universal joint was coupled with a prismatic joint. Finally, the universal joint was replaced by a spherical joint by Lenarcic and Stanistic (2003) (Figure 9). Alternatively, a robotic parallel mechanism has also been proposed by Lenarcic and Stanistic (2003). 15 .
(129) Correction methods and models available. Figure 9 : Different equivalent shoulder girdle models (Lenarčič and Stanišić, 2003) (adapted from (Sapio et al., 2006)). This latest model is a perfect example of equivalent model. With this robotic model, it is po possible ve a precise positioning of the GH joint. However, it remains difficult to reobtain the p to have proper matics of each bone of the shoulder girdle with such an approach. Consequently, all these kinematics. models ls allow a simple representation of the phenomenon occurring in the shoulder girdle but b do not allow llow a precise understanding of the different bones kinematics. They were mainly used us as simplification Yang et ification for musculoskeletal models or for ergonomic studies (Malek et al., 2006; Ya al., 2005) approach 005) where the aim is to determine the reachable space of the upper limb. This app seemss thus limiting as the different scapular girdle structures cannot be studied separately separately.. 2.1.2.2.Anatomical 2.1.2.2. Anatomical representation: Open loop vs Closed loo loop Open n loop model: Considering the different bony structures of the shoulder gir girdle In an open loop mechanism (Figure 10), all the different bony structures of the shoulder girdle are considered. onsidered.. The thorax, the clavicle and the scapula are connected in pair by single. mechanical anical links, mostly spherical joints (Engin and Tümer, 1989; Högfors et al., 1991; Yang Ya et 010). al., 2010).. . Figure 10 : Example of an open-loop mechanism. Consequently, it is possible to obtain the complete kinematics of the different body segments, with a representation close to the anatomy. However, compared to the shoulder girdle equivalent 16 .
(130) Correction methods and models available mechanism which integrates indirectly the scapulothoracic constraints, this open loop model eludes it completely, which seems less physiological. Even if these models are easy to implement due to their simplicity, it could be interesting to consider the ST joint through the use of a closed loop model. Closed loop model: Integration of the scapulothoracic joint The closed loop model takes into account the ST joint controlling the scapula position relative to the thorax. The ST joint consists in a sliding of the scapula between different muscular and bursal planes. Consequently, the model allows representing a sliding movement composed of different rotation and translation movements. Therefore, the classical mechanical links are not used to represent this specific joint and are replaced by geometrical constraints. Commonly in the literature, this joint is considered as a contact between an ellipsoid representing the thorax and different points of the scapula (Maurel, 1995; Tondu, 2005) (Figure 11).. Figure 11: Shoulder girdle model considering scapulothoracic joint (Maurel, 1995) in (Yang et al., 2010). Different models allowing three, four or five DoF have been proposed (Hill et al., 2008; Tondu, 2007) differing only by the number of contact points between the ellipsoid and the scapula (Figure 12). Thus, the three DoF model allows 2 translations and 1 rotation thanks to three contact points (Figure 12-c). By removing one contact point (Figure 12-b), one rotation is added. Finally, when only 1 point of contact (Figure12-a) is maintained, 2 translations and 3 rotations are allowed (Yang, 2003).. Figure 12: Different contact models of the scapulothoracic joint (Maurel, 1995) in (Yang et al., 2010). 17 .
Figure
+7
Documents relatifs
